NOT ONLY scrap and mathematics, chips and algorithms behind the AI there is also so much poetry. Literally, in various forms. Although it is true that Google thinks that the timing for the man-machine interaction is ripe, and about important news announced at the annual conference for developers, it is also true that this dialogue is still very cumbersome. Too much. The examples are within earshot: just choose the wrong words and Siri, just to give an example well known, does not understand our request. We misunderstand the point that we need to rephrase the question, note the terms and tone of voice. Thus the interrupted dialogue. The mechanism jams. And then it is in coaching the natural language cognitive systems that are hidden behind the scenes of the company, the new chatbot, like the smart screen readers, which are focusing the efforts of the big dell ‘ hi-tech. Not an easy feat, became nullifying the entire Silicon Valley. Which involves different dynamics and even different players. “The next job hot in the United States is for poets,” headlined in this respect the Washington Post a few weeks ago. The reference is to a new phenomenon that is catching on in the US innovation basin: professional communicators who are recruited by the giants of the world of technology, such as large and small startups are working on developing an artificial intelligence increasingly based on deep learning, or of the technology that mimics the behavior of our neurons. The ultimate objective of the call to arms is to make interactions with devices that offer us more conversational as possible. Moreover, given that the technological differences gradually become thinner, up to flatten, it is the experience that seems to play the game of the future. A pc that, right now, rather than look like a human, just wants to seem human. And always to reach destination in Google home are also along another path. Paved with 2,865 romance novels. So many in fact are the books that Big G has fed to one of its machine learning systems. To educate the art of conversation. The immediate result, and accidentally, was a computer with the intention of prophet: capable of rattling of similar propositions in their poems, given an initial and a final sentence. What future we’ll find out soon. And it may have implications even on Google Home, digital aide table, just presented. That might give us super custom responses. “In our app answers are proactive,” he confessed bitterly BuzzFeed News Andrew From , the Mountain engineer View that is working on the project, together with his colleague Oriol Vynials . “With this work, and upcoming activities, we hope to make the experience more like a real conversation, or to have different tones, styles and registers”. “Cognitive systems are able to read, learn, understand and respond,” reminds us Alessandro Curioni , director of the Zurich Research Lab headed by IBM, the company owns one of the brains made more chips interesting in circulation at the time. “And we have already demonstrated at various levels with Watson. What, for example, it was fed thousands of cases and legal documents. On them the machine can be based to assist lawyers by finding the right arguments for or against , of a case. and in this way help them to win the case. so I say: read books is simple, but when we speak of the ability to understand their message and to articulate an opinion here that things become really interesting. ” The transition from form to substance.
writes journalist Elizabeth Dwoskin :” Behind Apple’s Siri, the Amazon and Alexa Cortana Microsoft no longer just software engineers. Rather an increasing number of poets, comedians, writers fiction and other artistic figures, responsible for developing the personalities of these artificial intelligence tools for rapid growth. ” An idea that has its own foundation. “The natural language processing is not easy. But if we want to create an identity that speaks well and is equipped with a certain type of dialogue, so that it agreed better to those who use it, the communication experts by hiring the right people, “he says convinced Luigi Portinale , professor of computer science and artificial intelligence at the University of Eastern Piedmont. “For example, could intervene in the construction mode of response sentence, choosing from a more erudite vocabulary and another more colloquial, depending on the target audience to which it addresses the instrument at the moment, whether it is a young or of an adult “. Because words are important. Especially if the other part of the screen there is a computer speaker.
Annabell, the artificial neural network that has learned to speak
And in this direction, according to Francesco Lentini , Eloisa dad, chatbot “endowed with artificial personality” was born back in 1992, when he was still in Italy was not even internet, ruminating novels is of little value. “We can make him read all articles and essays in the world,” he says, “but we will not get a smarter computer, but what the French call idiot savant, the wise fool. Instead, I think the right approach is to work with the experience: why my Eloisa speaks with humans worldwide from February of ’92 and has since responded to over 100 thousand applications, improving question after question. “
Annabell, a simulated model of the pC used to represent processes that occur in our mind, seems to agree with him. Tells Wired.it : This scientific work (and not commercial like that of Lentini) shows that artificial neural networks can learn how to communicate through language, without any pre-codified language proficiency. Starting from scratch. A good result. Although his vocabulary is not very rich: similar to that of a child. Lentini concludes: “When they tell me: ‘We are trying to make the most human computer’, I reply that it’s no use. We need to create a true artificial intelligence, who does not think like us and gives responses that we ourselves would not be able to give. that means that it can help us. ” He’s working on it. And we hear some good.
No comments:
Post a Comment